SQL queries sometimes put floats in GROUP BY. E.g. if you have a many-to-one relationship you might do a query like
SELECT foo_id, foo.some_float, SUM(bar.some_thing)
FROM foo JOIN bar USING (foo_id)
GROUP BY foo_id, foo.some_float
I feel kinda dirty whenever I do this.
Though, I would guess the optimizer (at least in postgres) is smart enough to ensure no float equality checks actually happen under-the-hood. They could be necessary, if the schema was different than I'm imagining; but maybe in that case, it would almost always be a bad idea.
Though, I would guess the optimizer (at least in postgres) is smart enough to ensure no float equality checks actually happen under-the-hood. They could be necessary, if the schema was different than I'm imagining; but maybe in that case, it would almost always be a bad idea.